Anthropic's November listing will test what public investors pay for frontier-lab losses

Commitments already on Anthropic's books set how much capital it needs. Reuters obtained what appears to be a draft prospectus ^[2]. It shows $518 billion committed to cloud, computing and infrastructure ^[4], about 113 times the $4.6 billion of revenue Anthropic reported last year ^[2]. An operating loss of around $8 billion on that revenue ^[3] implies roughly $12.6 billion of operating costs, or about $2.74 spent for each dollar earned ^[1]. The commitments run over the coming years, but the financing decision falls this quarter, with trading now targeted for November ^[8].

The Newcomer newsletter does not think waiting for a calmer window is a real option. When a company needs as much capital as the frontier labs do, it wrote, "at some point you have to plug your nose and jump" ^[14].

The newsletter also makes the strongest objection to reading too much into the draft. Every figure in it predates Opus 4.6, the February 2026 model the newsletter credits with Anthropic's lead in agentic coding and a surge in revenue ^[5]. "Judging Anthropic without Opus 4.6 would be like evaluating Marvel before Iron Man," it wrote ^[6]. That objection applies to revenue. Growth since February would shrink the ratio of commitments to revenue, but it would leave the $518 billion commitment in place ^[4].

Growth decides the valuation. The newsletter reports open questions about Anthropic's growth rate and about whether it can reach $2 trillion at listing ^[7]. That price would be about 435 times last year's reported revenue ^[3]. OpenAI's private target of $1.4 trillion is about 20 times the annualized revenue of nearly $70 billion that several outlets reported this week ^[11][13][4]. The two ratios are not like for like, because one uses a past year's revenue and the other a current run-rate. At OpenAI's multiple, a $2 trillion Anthropic would need annualized revenue of about $100 billion ^[5].

OpenAI has put its financing in the opposite order. It is raising private money to last until 2027 ^[11], and Sam Altman has described the delay as a response to concerns about model safety ^[12]. Anthropic gets public capital in November and, in return, accepts a share price that moves every trading day. I'd expect OpenAI's eventual prospectus to be read against whatever price Anthropic sets.

For companies building on these models, the IPO matters because their vendors depend on outside funding. Both labs are trying to land what the newsletter called "their next tranche of desperately needed capital" ^[16]. A vendor spending about $2.74 for each dollar of revenue ^[1] needs that capital to arrive. The sources do not say how a weak debut would show up in API prices or contract terms. The newsletter expects the cycle to run longer across the industry if Anthropic's full financials show revenue still booming ^[15]. "If we start to see a slowdown, then things will be rough for the foreseeable future," it wrote ^[15].